© 2004 AlphaMed Press Maximizing the Potential of Bevacizumab in Cancer Treatmenta Department of Medicine, University of California, San Francisco, California, USA; b Memorial Sloan-Kettering Cancer Center, New York, New York, USA Correspondence: Emily Bergsland, M.D., Department of Medicine, University of California, San Francisco, 1600 Divisidero Street, 4th Floor, San Francisco, California 94115, USA. Telephone: 415-353-9888; Fax: 415-353-9959; e-mail: emilyb{at}medicine.ucsf.edu
Access and take the CME test online and receive one hour of AMA PRA category 1 credit atCME.TheOncologist.com
Promising results have been obtained with bevacizumab (AvestinTM; Genentech, Inc.; South San Francisco, CA) in clinical trials in patients with a range of solid tumors; however, to maximize the potential of this agent, further research is needed to clarify a number of important issues. These include the optimization of bevacizumab dosage and schedule of administration, the potential value of this agent in combination with other treatment modalities like chemotherapy and radiation, the management of toxicities, and the selection of patients most likely to benefit from treatment. Intriguing results from two recent phase III trials highlight the need for a better understanding of the best ways to incorporate bevacizumab into clinical practice. Ultimately, maximizing the potential value of this agent may require a more thorough understanding of bevacizumabs mechanism of action and the pathways mediating resistance. Key Words. Solid tumors • Bevacizumab • Chemotherapy • Clinical trials
The anti-vascular endothelial growth factor (VEGF) monoclonal antibody bevacizumab (AvastinTM; Genentech, Inc.; South San Francisco, CA) is currently in phase II/III trials in a wide range of tumor types, and promising results have been obtained in colorectal cancer [1, 2], non-small cell lung cancer (NSCLC) [3], and renal cell cancer [4]; however, there are a number of issues that have yet to be addressed in order to maximize the clinical potential of bevacizumab. These include the optimization of bevacizumab dosage and schedule of administration, the exploration of the potential value of this agent in combination with other treatment modalities, the management of toxicities, and the selection of patients most likely to benefit from treatment.
Defining the optimum dosage and schedule for antiangiogenic agents has proven to be a challenge. Traditionally, maximal tumor-cell kill was thought to be achieved by using the maximum tolerated dose (MTD) of a chemotherapeutic agent. The cytotoxic agent is typically given in a cyclic fashion, with scheduled breaks between treatments to allow for recovery of nontarget organs and tissues. However, such an approach is unlikely to be ideal for antiangiogenic agents, which may require long-term continuous administration for maximum efficacy. Furthermore, the relationship between the MTD and the optimal biologically active dose (OBD) is unclear for antiangiogenic agents. For some agents, the MTD may greatly exceed the OBD required to maximally block angiogenesis; moreover, for some anti-VEGF or VEGF-receptor agents, no dose-limiting toxicities have been observed, further complicating our efforts to define the optimal dose. Finally, it is conceivable that the optimum dosage may vary with tumor type, location, and growth rate and with type of prior therapy.
Studies in preclinical models predicted that it would be necessary to have trough plasma bevacizumab concentrations of 1030 µg/ml in order to achieve maximal tumor growth inhibition [5]. The terminal elimination half-life of bevacizumab is long (12 weeks) in all species, and 125iodine antibody localization studies indicate that it is distributed to highly perfused areas (albeit with minimal localization to the liver) [5, 6]. Doses of 0.110 mg/kg/week were evaluated in phase I clinical trials with bevacizumab and indicated that bevacizumab has a linear pharmacokinetic profile. Doses
The dose levels explored in phase II clinical trials with bevacizumab ranged from 320 mg/kg administered every 2 or 3 weeks to patients with various advanced solid organ tumors. Interestingly, a consistent dose-response relationship was not observed in all trials. In metastatic NSCLC and metastatic renal cell carcinoma, a traditional dose-response relationship was observed; higher doses of bevacizumab appeared to be more efficacious than lower doses of the drug. Bevacizumab at 15 mg/kg appeared to be more effective than bevacizumab at 7.5 mg/kg when added to carboplatin/paclitaxel chemotherapy in metastatic NSCLC [3], and 10 mg/kg bevacizumab was more effective as a single agent than 3 mg/kg bevacizumab in metastatic renal cell carcinoma [4]. In contrast, the dose-response relationship for bevacizumab was unclear in the phase II breast cancer and colorectal cancer clinical trials (Table 1
The roles of loading dose and intrapatient dose escalation also remain to be clarified. A loading dose was used in the renal cell cancer study [4] and may be particularly important when time to progression is used as the primary end point (i.e., when rapid achievement of therapeutic levels is essential). Questions regarding optimal bevacizumab dose and schedule are not likely to be answered immediately; doses of 515 mg/kg delivered every 23 weeks are under investigation in phase III trials, but none of the ongoing studies are designed to directly address this issue.
In contrast to traditional therapies, which can result in rapid tumor regression, assessing clinical activity of angiogenesis inhibitors may be problematic, as some agents may take as long as 1 year to induce tumor regression [8], and there are no validated surrogate markers of biological activity. The purported mechanism of action of bevacizumab and the potential for delayed efficacy had led to speculation that time to progression or overall survival may be more relevant measures of activity than objective response rate. Interestingly, improvements in objective response rate and time to progression translated into better overall survival in patients with metastatic colorectal cancer receiving first-line chemotherapy plus bevacizumab [1, 2]. In contrast, an improvement in the objective response rate did not ultimately lead to longer overall survival in previously treated patients with metastatic breast cancer assigned to receive bevacizumab plus capecitabine compared with chemotherapy alone [9]. Furthermore, the significantly longer time to progression observed in patients with renal cell cancer treated with bevacizumab alone did not translate into longer overall survival [4]. These results suggest that the optimal end point for evaluating response to therapy may need to be selected in the context of prior therapy and may prove to be stage and/or disease dependent. Reliable surrogate markers of activity are needed to quantify early changes induced by these agents. Efforts are being made to adapt techniques such as magnetic resonance imaging, computerized tomography, positron emission tomography, and ultrasound [10] as tools for assessing early evidence of antiangiogenic activity. The use of serial tumor biopsies as well as serum or urine concentrations of angiogenic factors are also under investigation. In addition to indicating response to therapy, biological markers may also help to identify which individuals are most likely to benefit from antiangiogenic therapy. Measurements of circulating levels of VEGF and basic fibroblast growth factor have proven helpful in some instances, but the plethora of angiogenic factors involved in tumor-associated angiogenesis implies that relying on any single angiogenic factor may be impractical and misleading [11, 12]. Developing preclinical models that more accurately predict the clinical activity of angiogenesis inhibitors is also a priority. Most investigators rely on mouse models that involve the use of rapidly growing transplantable murine tumors or human tumor xenografts implanted subcutaneously. These models may not accurately replicate the behavior of a tumor in its normal site of origin. Furthermore, the ability of these models to accurately predict clinical efficacy may be limited by the fact that the pattern of metastatic spread that is typically seen clinically in patients may not be faithfully recapitulated in the tumor-bearing animals. With respect to antiangiogenic agents specifically, the features of tumor-associated angiogenesis may be organ or site specific, further impacting the utility of subcutaneous transplant models. While validation will be required, testing angiogenesis inhibitors in animals that model bulky metastatic deposits to the organ sites commonly involved in humans (e.g., lung, liver, brain, and bone) may be more predictive of clinical efficacy. To this end, orthotopic, rather than heterotopic, transplant models may be superior to subcutaneous xenograft models. Genetically engineered transgenic mouse models in which immunocompetent mice spontaneously develop tumors may prove particularly informative [13, 14].
Chemotherapy Bevacizumab is being evaluated in combination with a range of different chemotherapies; however, the optimal drug combinations, dose, and sequence of administration have not yet been defined. In an effort to maximize the inhibitory effects of chemotherapy on endothelial cells, it has been suggested that the use of smaller and more frequent doses without scheduled breaks (metronomic chemotherapy) may be a superior means to inhibit endothelial cell growth than the traditional high-dose pulsed regimens [13, 15, 16]. This strategy may also help to limit toxicity. Most common chemotherapeutic agents are able to inhibit angiogenesis. However, since tumor-associated endothelial cells proliferate at a slower rate than tumor cells, their replication is only weakly disrupted by regimens that involve intermittent administration of relatively high-dose chemotherapy. Furthermore, the planned breaks between therapy with traditional dosing schedules allow cells to recover between cycles. In addition to direct effects on cultured endothelial cells, a number of cytotoxic drugs demonstrate antiangiogenic effects in preclinical animal models when delivered in a metronomic fashion. Experiments in murine models suggest that the addition of a dedicated antiangiogenic agent to a metronomic chemotherapy regimen potentiates antitumor activity [15, 17]. A study evaluating bevacizumab in combination with low-dose oral cyclophosphamide in patients with ovarian cancer is in progress. Another unresolved issue is the optimal sequence of treatment when antiangiogenic therapy is combined with chemotherapy. Antiangiogenic therapy may initially normalize the typically structurally and functionally abnormal tumor vasculature, thereby improving delivery of oxygen and chemotherapy. Optimal scheduling with chemotherapy could theoretically take advantage of this window of opportunity and allow cytotoxic therapy maximal access to tumor cells [11]. However, one also needs to consider the possibility that antiangiogenic agents may ultimately interfere with the delivery or activity of cytotoxic chemotherapy or other agents owing to their inhibitory action on the vasculature.
Other Targeted Therapies Bevacizumab is also being studied in combination with dendritic cell treatment. The rationale for evaluating this combination is based on clinical and preclinical data suggesting that VEGF has an inhibitory action on the maturation of antigen-presenting cells. Other approaches of interest for future studies include combining anti-VEGF therapy with inhibitors of the Akt and Raf kinase pathways.
Toxicity To date, no significant infusion-related symptoms have been noted in phase II trials. Hypertension and proteinuria were seen in all phase II studies with bevacizumab, with epistaxis and headache also reported. Headache, associated with nausea and vomiting, was considered to be dose limiting in patients with metastatic breast cancer receiving bevacizumab at a dose of 20 mg/kg [8]. The majority of patients who developed new or increased proteinuria in phase II trials was asymptomatic [1, 5]. Proteinuria has not been associated with evidence of renal dysfunction. While the mechanism of proteinuria has not yet been elucidated, the predominance of albumin in the urine and the presence of membranoproliferative glomerulonephritis in some patients [5] indicate that the site of activity is most likely the glomerulus. VEGF is expressed in the glomerulus, and glomerular endothelial repair is believed to be mediated through VEGF [18]. In addition, it is conceivable that low levels of erythropoietin in cancer patients may exacerbate the situation because erythropoietin stimulates VEGF release in the glomerulus [19]. Moreover, patients with more severe proteinuria (nephrotic syndrome) appeared to be more likely to have associated hypertension (both pre-existing and bevacizumab induced) [5]. Interestingly, recently reported phase III data in colorectal cancer suggest that proteinuria is not increased in patients receiving bevacizumab when patients with significant baseline proteinuria are excluded [2]. Hypertension has been reported in all studies involving bevacizumab. Although 84% of the cases in the phase II studies were grade 3/4, these were easily managed with antihypertensive therapy; bevacizumab was discontinued because of hypertension in only two patients [5]. The mechanism underlying bevacizumab-related hypertension is not yet clearly understood. Infusion of VEGF has been found to produce hypotension [20], and thus, blockade of VEGF may potentially lead to elevation of blood pressure. VEGF receptor blockade results in decreased production of the vasodilator nitric oxide (NO) [21]. Reduction of NO formation could also lead to reduced renal sodium excretion, which is known to be associated with persistent hypertension [22]. The most serious safety issue was identified in the bevacizumab NSCLC phase II study, in which six patients (9%) developed life-threatening episodes of pulmonary hemorrhage, four of which were fatal. These occurred in patients with centrally located cavitary lesions that were necrotic and predominantly associated with a squamous cell histology. Moreover, similar episodes of life-threatening bleeding have not been noted in trials of bevacizumab in patients with breast, prostate, or renal cell cancers [5]. Data from the phase II study in colorectal cancer suggest a possible association between bevacizumab and thromboembolic events in patients with metastatic disease [1]. Thromboembolic events were more common in both bevacizumab treatment groups (5 mg/kg and 10 mg/kg) than in the control group (nine, four, and three patients, respectively). Intriguing results from the recent phase III study in previously untreated patients with metastatic colorectal cancer, however, suggest that treatment with bevacizumab (at a dose of 5 mg/kg every 2 weeks) in combination with chemotherapy is not associated with a higher risk for thromboembolic events in this patient population [2]. Overall, an interim analysis of approximately 1,000 patients accrued in the National Cancer Institutes bevacizumab Cancer Therapy Evaluation Program collected 280 reports of serious adverse events (SAEs) to date [5]. Approximately 20% of those were judged to be "possibly related" to bevacizumab together with other concomitant factors. Approximately 10% were judged to be "possibly related" to bevacizumab alone and 15 SAE reports have been filed with the U.S. Food and Drug Administration. In addition to instances of hemorrhage, hypertension, and thromboembolism, other possible SAEs include instances of myocardial infarction in patients with coronary artery disease at baseline, reductions in left ventricular ejection fraction in patients with prior anthracycline therapy, pericardial effusion, and arrhythmias [5].
The traditional paradigm for developing new cancer therapies is rooted in disease-specific clinical trials that target clinically defined populations. In this model, essentially all patients within a given subgroup are eligible for inclusion in a pivotal clinical trial. The availability of novel, targeted, biologically based, therapeutic strategies has provided opportunities to more precisely define the population to be evaluated. For each new agent, it should be feasible to enrich the target patient population with those individuals most likely to respond to that agent. In theory, this capability should translate into a rational approach to anticancer therapy, leading to enhanced efficacy, a reduction in the number of patients required for statistically informative clinical trials, and the potential for more rapid integration of promising novel agents into clinical practice. While screening may limit enrollment, the evaluation of a targeted therapy such as bevacizumab in an unselected population may prove problematic because treatment effects could be diluted if only a subgroup of patients is likely to respond. Prolonged stabilization of disease lasting more than 3 years has been achieved with bevacizumab in a small fraction of patients with advanced solid tumors [23]. Understanding the particular tumor biology involved in these cases would be extremely helpful in guiding patient selection in the future. However, despite intensive research, validated biologic markers predictive of response to bevacizumab have not been identified, and screening for VEGF expression or the presence of cognate receptors is not routinely performed. VEGF expression is not ubiquitous, although it is present in 30%-60% of most solid tumors and in up to 100% of renal cell carcinomas. Expression of VEGF may be site or organ specific [24], and tumor-associated changes may be relatively small [25]. Nevertheless, small changes in VEGF expression may ultimately have profound effects on prognosis. Perhaps at least some of the trials involving bevacizumab should include a prospective assessment of VEGF levels and a requirement for a certain level of VEGF expression as a criterion for enrollment in the study. Further insight into this issue should result from retrospective correlative studies planned for several phase II/III trials aimed at identifying biologic markers that predict the response to bevacizumab. Interesting results from one of these correlative studies were recently presented. Archived primary tumor samples were obtained from previously treated patients with advanced breast cancer participating in the phase III trial designed to assess the activity of capecitabine with or without bevacizumab [9, 26]. VEGF expression was assessed by in situ hybridization, but response did not correlate with the level of VEGF expression. The value of the study was potentially limited by the relatively small sample size, the low objective response rate in the test group, and the use of primary tumor material to predict response in the face of metastatic disease. Nevertheless, the data are intriguing and underscore the potential complexities related to predicting response to bevacizumab. A number of questions remain related to the role of screening in patients treated with bevacizumab. One needs to consider the value of fresh versus archived tissue and the accessibility of biopsy material from primary tumors versus metastatic sites. The optimal assay for assessing VEGF or receptor expression (protein versus RNA levels) has not yet been identified, and defining a relevant level of expression remains a challenge. Also, the utility of testing for urine, serum, or plasma VEGF is unknown. Finally, it is important to note that expression of VEGF does not guarantee that VEGF is a valid target in a particular patient. Screening for other members of the VEGF family, besides VEGF-A, or assessing the expression or activation status of VEGF receptors or downstream markers of VEGF activity may prove more fruitful. The fact that metastatic disease is often marked by the production of multiple mediators of angiogenesis may prove to be a potential obstacle to the successful use of a therapy solely targeting VEGF [27]. Thus, a selective approach to the use of bevacizumab may be warranted. For example, the progression of breast cancer involves the contribution of other proangiogenic factors, and there is heterogeneity of VEGF expression. Bevacizumab may be more effective in the early stages of breast cancer, when VEGF may be the predominant proangiogenic factor secreted. Moreover, large tumors may produce very large amounts of angiogenic factors, which, theoretically, may overwhelm antiangiogenic agents. Since the initial clinical testing of novel antiangiogenic agents is traditionally conducted in heavily pretreated patients with advanced and bulky disease, one has to consider the possibility that treating an unscreened population may lead one to underestimate the potential value of a given agent. Further investigation is required to define when and in which patients antiangiogenic agents like bevacizumab should be given in order to maximize their efficacy.
The advent of antiangiogenic therapy has left physicians faced with a number of challenges as well as opportunities for improving anticancer treatments. A range of issues in relation to the optimal use of bevacizumab has yet to be clarified, but preclinical data are lacking in many areas. The question of how to optimize the dosage and schedule of an agent without a consistent dose-response relationship and without a validated biologic marker of activity has yet to be answered, and the roles of loading doses and dose escalation remain unclear. Recent phase III data suggest that VEGF is a valid therapeutic target in some patient populations. However, although bevacizumab lacks the typical adverse events of chemotherapy, the management of toxicities requires further study. Furthermore, the optimal way to combine and sequence bevacizumab with other treatment modalities remains to be elucidated, just as biologic markers predictive of response have yet to be identified. Research is ongoing to address these issues, and significant advances are likely to arise from a better understanding of bevacizumabs mechanism of action and the pathways mediating resistance to the drug.
E.B. is a consultant for Genentech. M.N.D. receives research support and honoraria from Genentech.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||